Detection of common risk factors for diagnosis of cardiac arrhythmia using machine learning algorithm

This article aims to establish an accurate and innovative objective framework for classification of cardiac arrhythmia patients by trying to measure the importance of specific factors that are potentially relevant to its diagnosis. Cardiac arrhythmia (CA) is a group of condition related to the irregular heartbeats. It is very essential to prevent a CAs, as they are the most common cause of natural death in all over the world. According to the health reports, more than 4.5 lakh cardiac patients fatalities annually in the United States alone. To diagnose cardiac diseases, patient’s reported qualitative symptoms can be useful. However, this strategy may fail sometimes due to less accuracy and false positive cases. Therefore in this work, we strive to find a quantitative basis for more reliable and accurate diagnosis of cardiac arrhythmias. This research used the openly available MIMIC-III database to obtain large quantities of clinical monitoring data from patients over the age of sixteen admitted to intensive care units (ICUs). The database was processed on the Health Sciences and Technology (HEST) Cluster, filtered with in a specified time frame(24hrs, 12hrs and 6hrs) and organized into a multi-class and a single-class and finally split into train, validation, and test sets with respective weights of 0.7, 0.2, and 0.1. We used random forest classifier model for the diagnosis of cardiac arrhythmia and measure the importance of different features like respiratory rate, blood pressure, sodium, potassium, calcium, among the other features. Hyperparameter optimization techniques like grid search and genetic algorithms are compared to find the maximum number and depth of trees in the forest. The model achieved, at its best, an Area Under the Receiver Operator Curve (AUC) score of 0.9787 and, thus, confirmed the importance of several previously suggested factors in the diagnosis of cardiac arrhythmias. We substantiated claims that each of sodium, calcium, potassium, respiratory rates and blood pressure can be used for the early diagnosis of cardiac arrhythmias.